Opinion

Felix Salmon

The negative correlation between obesity and indebtedness

By Felix Salmon
October 2, 2011

Michael Lewis says something very odd in his big piece on California and the phenomenon of overconsumption:

The succession of financial bubbles, and the amassing of personal and public debt, Whybrow views as simply an expression of the lizard-brained way of life. A color-coded map of American personal indebtedness could be laid on top of the Centers for Disease Control’s color-coded map that illustrates the fantastic rise in rates of obesity across the United States since 1985 without disturbing the general pattern.

map26.jpg

Here is the map in question; if you go to the site and see it animated over time, it is indeed quite scary. But it doesn’t look remotely like a map of American personal indebtedness.

Indeed, if you download the Fed’s list of total debt balance per capita, by state, it looks nothing like this map at all. The Fed only lists the ten biggest states, and the overall average, which is is a per capita indebtedness of $47,260.

At the bottom of the personal-indebtedness league table are Texas ($34,640 of debt, 31% obesity) and Ohio ($34,090 of debt, 29.2% obesity).

Meanwhile, the states at the top of the personal-indebtedness league table are California ($73,300 of debt, 24% obesity), New Jersey ($60,560 of debt, 23.8% obesity), and Nevada ($60,190 of debt, 22.4% obesity).

In fact, indebtedness and obesity have a strong negative correlation. If I plug the obesity rates for the ten largest states into an online correlation calculator, I get an amazing -0.843 correlation between obesity rates and personal indebtedness. What Lewis wrote is so false that the opposite of it is actually true.

Now there are strong connections between debt and obesity. For one thing, both have been increasing steadily over time. And a German study last year showed that over-indebted Germans were more than twice as likely to be obese as their financially-successful counterparts. But here in the US, Texas, with its responsible lenders and no housing bubble, is much fatter than places like California and Nevada, which went on bubble-fueled binges of borrowing and irresponsible lending.

Comments
17 comments so far | RSS Comments RSS

High obesity is prevalent in the poorer states. If you measure in constant dollars, there just isn’t enough money in the poorer states to show the correlation between obesity and debt. If you measure debt as a percentage of median income or or median net worth, you’ll find your correlation.

Posted by mkvc | Report as abusive
 

nice find.

Lewis’s language is imprecise in that it’s hard to tell if he’s referring to the current levels of obesity and personal debt, or if he wants to talk about the changes in those variables over time.

It is possible (and likely) that when you look at the change in obesity for states between 1985 and 2010 and the change in debt for states over the same time period you’ll find a positive correlation — i.e. both obesity and indebtedness have risen since 1985. what a shocker.

(but state-level personal debt data going back that far probably isn’t easy to come by, so this is conveniently untestable.)

Posted by zubin | Report as abusive
 

Similar to mkvc, my knee-jerk reaction was that you ought to control for some measure of personal income. But actually running the analysis in the cross-section using 2010 data suggests Felix’s result is robust to using state median income as a control. I get a sample correlation coefficient of -0.75.

Using the available data, you could try a time series analysis from 1999-2010 using the 10-state subsample.

Posted by nmph | Report as abusive
 

Similar to mkvc, my knee-jerk reaction was that you ought to control for some measure of personal income. But actually running the analysis in the cross-section using 2010 data suggests Felix’s result is robust to using state median income as a control. I get a sample correlation coefficient of -0.75.

Using the available data, you could try a time series analysis from 1999-2010 using the 10-state subsample.

Posted by nmph | Report as abusive
 

as per nmph, I looked at correlation between changes in obesity and % change in indebtedness for 1999 through 2010 and it matches felix’s result in direction (corr = -0.51). While obesity and debt did both increase, the increase in debt was higher for the states w/ smaller increases in obesity.

And it doesn’t look like a base effect is driving the result. The correlation for the seven states with roughly the same percentage obese in 1999 is even stronger: -.87

interestingly, this german study did not find a causal effect of indebtedness on obesity: http://goo.gl/TbE0X

Posted by zubin | Report as abusive
 

Perhaps I shouldn’t question the science behind a politically-motivated study, but why the heck should obesity and debt have anything to do with one another? Have you studied risk factors for obesity? Have you studied risk factors for debt? Is there any reason they SHOULD be connected?

Moreover, running a correlation on state-level averages is pretty silly. If you want to show (or deny) a connection, then why not work with individual data? Seems to me to be bad science all around. Reason enough to ignore the “study”.

That said, indebtedness has a curious relationship to income. If you don’t have any income, you won’t find many people willing to loan you money. The poor typically have less debt (proportionate or otherwise) than the middle-class.

Posted by TFF | Report as abusive
 

Indebtedness (at the moment) has much more to do with housing prices than income.

Posted by mattmc | Report as abusive
 

@TFF, Lewis made the comparison because he was talking about how humans are wired to take whatever they can get, it wasn’t a politically motivated study (it wasn’t a study at all). People become obese because they eat all that is available to them, whether they need it or not, regardless of the long term consequences. The debt crisis occurred, it is proposed, because people (individually and collectively) assumed as much debt as was available to them, whether they needed it or not, regardless of the long-term consequences. Lewis suggested that obesity increased at the same rates as personal debt, not that the personal debt was proportional to obesity, because it is the same innate behavior that causes both problems.

I don’t think he is suggesting the lizard core of the human brain is responsible for the debt crisis, but rather just trying to explain how so many people and the nation could make such unhealthy choices.

Posted by KenG_CA | Report as abusive
 

In the US obesity is negatively correlated with income, and real estate is more expensive in high income states. There was a lot more collateral value in the more expensive states and people with more money tend to have higher FICO scores, so you’d expect a lot more debt in the high income, low obesity states.

Posted by Kaleberg | Report as abusive
 

The fact that obesity and debt don’t in fact correlate is unsurprising if we look at the actual causes of the obesity epidemic. Lack of self-discipline has little to do with it.

People to a large extent are not getting fatter because they eating to much. They are getting fat because the foods they are eating are more fattening. The rise in obesity rates correlates with a rise in the consumption of refined carbohydrate. Because of its effect on metabolism on a majority of the population refined carbohydrate is insidiously obsesogenic.

Endocrinology tells us that this is the case, but economics reinforces the thesis. We are all told that the poor are more vulnerable to weight gain because the cheapest food is the most calorie dense. Junk food provides more calories per dollar. But why would someone with scarce dollars continue to purchase food after their caloric needs have been met?

The fact is, someone with insulin resistance is overeating because they are getting fat rather than the other way around. They eat potato chips, their muscle tissue no longer responds to insulin signals to accept glucose for energy. High blood sugar causes the insulin to signal the adipose tissue to store the energy as fat. The muscles still need energy signaling the brain to seek food. Rinse and repeat.

This obviously is a simplification, leptin and ghrelin play reinforcing roles; not everyone who is overweight is insulin resistant, but it describes the problem for most of the population.

I apologize for the poorly written tangential rant, tired canard that we are getting fat because we eat to much needs a little light shined on it. Eating too much describes HOW we gain weight. It doesn’t describe WHY we are gaining weight. Food availability doesn’t tell the whole story. Affluent people have MORE food available than poor people. If it was simple as availability the correlation between weight gain and income would be reversed.

Posted by MarcBrazeau | Report as abusive
 

Some good comments in this thread.

I enjoy reading Lewis – he weaves a good story, but this one felt a bit forced. The whole ‘lizard brain’ bit is well-trod ground, and trying to wrap up indebtedness and obesity as correlated moral failings without checking the data is lazy. The best part of his writing is usually his ability to empathetically portray people with telling details. For instance, the bit about Schwarzenegger riding the bike was great. The broader ‘tragedy of the commons’ thesis didn’t offer too much. Still, I’d rather read Lewis than almost anybody else on these kinds of subjects.

Posted by JimInMissoula | Report as abusive
 

Try it again using a county level map and it will fit much, much better.

Posted by SohoD | Report as abusive
 

“The rise in obesity rates correlates with a rise in the consumption of refined carbohydrate.”

Actually no, it correlates with a rise in consumption of processed foods in general, and a simultaneous collapse in physical activity. I know that it’s fashionable these days to blame refined carbs (and some people even extend it to all carbs, including fruit), and I’m not a big fan of refined carbs either, but the truth is, refined carbs (cane sugar) have been consumed in the Western world copiously since the late 1800′s. The obesity epidemic is a lot more recent. Perhaps because we didn’t have Pringles chips and Double Western Bacon Cheeseburgers. In Europe, the epidemic is only just starting in the last 10-20 years as American fast-food chains and snack manufacturers make inroads into their markets.

Posted by Nameless | Report as abusive
 

“We are all told that the poor are more vulnerable to weight gain because the cheapest food is the most calorie dense. Junk food provides more calories per dollar.”

It’s a bit more complicated than that. Junk food is the most calorie dense – yes. Palatable, calorie dense food messes with leptin pathways and leads to weight gain – yes. Junk food is cheaper than “normal” food – no.

Basic food ingredients are dirt cheap. You can buy two pounds of frozen chicken breast or four pounds of apples for the price of one Big Mac. It is not that expensive to eat healthy. BUT – you have to cook your food yourself. One of the strongest predictors of being overweight is the frequency of visits to fast-food restaurants. Lack of self-discipline manifests in making poor food choices (eating Pringles instead of apples), and in choosing to eat restaurant junk food because one is too lazy to cook.

Posted by Nameless | Report as abusive
 

“Basic food ingredients are dirt cheap. You can buy two pounds of frozen chicken breast or four pounds of apples for the price of one Big Mac. It is not that expensive to eat healthy. BUT – you have to cook your food yourself.”

Deserves to be repeated! Cook your own food from basic ingredients and you can easily (richly!) feed a family of four on $150/week. Eat at McDonalds regularly and you could spend twice that much.

Posted by TFF | Report as abusive
 

Back to the subject of debt. I would expect to see a correlation between obesity and lack of self-discipline. But indebtedness is not a marker of the lack of self-discipline. If anything, it’s the opposite. With notable exceptions, banks are not in the business of giving away megabuck loans to undisciplined people. In fact, poorest and fattest people can be expected not to have a mortgage, to have very short credit lines, and possibly to have a bankruptcy on their credit records.

And there is, in fact, a negative correlation between average credit scores and obesity rates. The states with worst credit scores are typically Texas, Louisiana, and Nevada. The best scores are in Minnesota, North/South Dakota, and Vermont.

Nevada is an outlier with low obesity and bad credit scores, but that may be the consequence of the property bubble. I tried to find data from 2003 or earlier, but didn’t have any luck.

Posted by Nameless | Report as abusive
 

One must be careful in using correlation statistics. There is a well known statistically significant correlation between ice cream sales in London and deaths in Bombay India. Interesting and strange, it does not mean that Londoners should eat less ice cream to save those in Bombay.

Posted by zotdoc | Report as abusive
 

Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/
  •